This work deals with the combined effect of nonlinear distortions and inter-channel interference in millimeter wave multi-input multi-output (MIMO) communications. Deep neural networks (DNNs) can be used to handle the effect, but they often require a large number of pilot symbols, hindering their applications. With the aim of online training using a relatively small number of pilot symbols, we design a deep neural network (DNN) architecture carefully, which consists of a fully connected linear hidden layer and a non-fully connected nonlinear hidden layer. The linear hidden layer is used to deal with the co-channel interference and the nonlinear hidden layer is used to handle the nonlinear distortions. Moreover, the parameters of the DNN are properly tied to reduce the number of independent parameters. With such a DNN, the receiver is much efficient in terms of training overhead and symbol error rate performance, compared to conventional (DNN-based) techniques. Simulation results demons
Abstract
Visible light communications (VLC) employ visible light for communication within the frequency spectrum from 430 THz to 790 THz. The ultra-wide bandwidth resolves the spectrum exhaustion in radio frequency communications. Light-fidelity (LiFi) takes VLC further to achieve high speed, bi-directional and fully networked wireless communications. Light-emitting diode (LED) is the major light source employed by LiFi as LED is energy efficient for illumination, and at the same time can achieve high speed data transmission thanks to the LED fast switching capability. Now, LED communications, one of the major application scenarios of LiFi, have attracted huge attention and research interests. A variety of advantages are offered including the unregulated wide bandwidth, energy efficiency, safety, etc., which makes LED communications a promising supplement to the fifth generation (5G) communications. LED is the major source of nonlinearity in LED communications and LED also exhibits